[jira] [Created] (FLINK-11874) [checkpoint] Split CheckpointStorage interface to distinguish JM and TM side

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

[jira] [Created] (FLINK-11874) [checkpoint] Split CheckpointStorage interface to distinguish JM and TM side

Shang Yuanchun (Jira)
Yun Tang created FLINK-11874:
--------------------------------

             Summary: [checkpoint] Split CheckpointStorage interface to distinguish JM and TM side
                 Key: FLINK-11874
                 URL: https://issues.apache.org/jira/browse/FLINK-11874
             Project: Flink
          Issue Type: Improvement
          Components: Runtime / Checkpointing
            Reporter: Yun Tang
            Assignee: Yun Tang
             Fix For: 1.9.0


Currently, interface {{CheckpointStorage}} mixed JM and TM side, which would be confusing for users and developers to distinguish them.

Take [FLINK-11696|https://issues.apache.org/jira/browse/FLINK-11696] as an example, the directories should only be created once from JM side. However, since we mixed the JM and TM side, TMs would also create directories again.

We could let interface {{CheckpointStorage}} only has two methods:

{code:java}
CheckpointStreamFactory resolveCheckpointStorageLocation(
                        long checkpointId,
                        CheckpointStorageLocationReference reference)

CheckpointStateOutputStream createTaskOwnedStateStream()
{code}

And a new interface {{ChekcpointCoordinatorStorgae}} could be introduced extending {{CheckpointStorage}} and have methods below:

{code:java}
boolean supportsHighlyAvailableStorage()

boolean hasDefaultSavepointLocation()

CompletedCheckpointStorageLocation resolveCheckpoint(String externalPointer)

CheckpointStorageLocation initializeLocationForCheckpoint(long checkpointId)

CheckpointStorageLocation initializeLocationForSavepoint(
                        long checkpointId,
                        @Nullable String externalLocationPointer)
{code}

With this refactor, JM would only use cast {{ChekcpointCoordinatorStorgae}} so that we could shield TMs from accidentally calling unexpected methods.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)